U.S. patent application number 14/095709 was filed with the patent office on 2015-06-04 for peak shaving using energy storage.
This patent application is currently assigned to SolarCity Corporation. The applicant listed for this patent is SolarCity Corporation. Invention is credited to Eric Daniel Carlson, Karthikeyan Varadarajan.
Application Number | 20150153394 14/095709 |
Document ID | / |
Family ID | 53265130 |
Filed Date | 2015-06-04 |
United States Patent
Application |
20150153394 |
Kind Code |
A1 |
Carlson; Eric Daniel ; et
al. |
June 4, 2015 |
PEAK SHAVING USING ENERGY STORAGE
Abstract
Techniques for controlling an energy storage device to reduce
peak power demand at a site are provided. In one embodiment,
instantaneous power usage at the site can be monitored, where the
instantaneous power usage corresponds to power that is
instantaneously imported or exported at a point of common coupling
(PCC) between the site and a utility-managed energy grid. A
historical power usage value for the site can then be calculated
based on the monitored instantaneous power usage, and the
historical power usage value can be compared with a target peak
value plus a buffer value. If the historical power usage value
exceeds the target peak value plus the buffer value, the target
peak value can be set to the historical power usage value.
Inventors: |
Carlson; Eric Daniel; (San
Mateo, CA) ; Varadarajan; Karthikeyan; (Mountain
View, CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
SolarCity Corporation |
San Mateo |
CA |
US |
|
|
Assignee: |
SolarCity Corporation
San Mateo
CA
|
Family ID: |
53265130 |
Appl. No.: |
14/095709 |
Filed: |
December 3, 2013 |
Current U.S.
Class: |
700/291 |
Current CPC
Class: |
H02J 3/32 20130101; G05B
15/02 20130101; G05B 2219/2642 20130101; G01R 21/1331 20130101;
H02J 3/30 20130101; H02J 3/28 20130101; Y04S 20/20 20130101; Y02B
70/30 20130101; H02J 7/35 20130101 |
International
Class: |
G01R 21/133 20060101
G01R021/133; G05B 15/02 20060101 G05B015/02 |
Claims
1. A method for controlling an energy storage device to reduce peak
power demand at a site, the method comprising: monitoring, by a
computer system, instantaneous power usage at the site, the
instantaneous power usage corresponding to power that is
instantaneously imported or exported at a point of common coupling
(PCC) between the site and a utility-managed energy grid;
calculating, by the computer system, a first historical power usage
value for the site based on the monitored instantaneous power
usage; comparing, by the computer system, the first historical
power usage value with a target peak value plus a buffer value; and
if the first historical power usage value exceeds the target peak
value plus the buffer value, setting, by the computer system, the
target peak value to the first historical power usage value.
2. The method of claim 1 wherein the first historical power usage
value is a maximum average of the monitored instantaneous power
usage over a rolling time window.
3. The method of claim 1 wherein the rolling time window is a time
interval that the utility uses to determine a peak demand charge
for the site.
4. The method of claim 3 wherein the rolling time window
corresponds to fifteen minutes.
5. The method of claim 1 wherein the buffer value is determined
empirically based on the site's historical power load.
6. The method of claim 1 further comprising calculating a maximum
allowable power output value for the energy storage device, wherein
the maximum allowable power output value decreases as a state of
charge of the energy storage device decreases.
7. The method of claim 6 wherein the maximum allowable power output
value is calculated using a function that takes as input a maximum
rated power of the energy storage device, a total amount of stored
energy in the energy storage device, and a target runtime for the
energy storage device.
8. The method of claim 7 wherein calculating the maximum allowable
power output value comprises taking the lesser of: the maximum
rated power; and the total amount of stored energy divided by the
target runtime.
9. The method of claim 6 further comprising calculating a second
historical power usage value for the site, the calculating of the
second historical power usage value comprising: for each of a
plurality of measurement time intervals of the instantaneous power
usage over an immediately preceding time window, computing an
average amount of power imported at the site for the measurement
time interval; and selecting the highest average amount across the
plurality of measurement time intervals.
10. The method of claim 9 further comprising, if the second
historical power usage value is greater than the target peak value,
discharging the energy storage device in accordance with either:
the maximum allowable power output value; or a power value that is
proportional to a maximum of, for each measurement time interval in
the plurality of measurement time intervals: a product of a time
range of the measurement interval and the difference between an
average demand for the time measurement interval and the target
peak value.
11. The method of claim 10 further comprising, if the second
historical power usage value is less than or equal to the target
peak value, discharging the energy storage device in accordance
with the lesser of: the maximum allowable power output value; and a
current power load of the site minus the target peak value.
12. The method of claim 11 wherein, if the current power load minus
the target peak value is a negative value, the energy storage
device is charged in accordance with the negative value.
13. A system comprising: a processor configured to: monitor
instantaneous power usage at the site, the instantaneous power
usage corresponding to power that is instantaneously imported or
exported at a PCC between the site and a utility-managed energy
grid; calculate a first historical power usage value for the site
based on the monitored instantaneous power usage; compare the first
historical power usage value with a target peak value plus a buffer
value; and if the first historical power usage value exceeds the
target peak value plus the buffer value, set the target peak value
to the first historical power usage value.
14. The system of claim 13 wherein the processor is further
configured to calculate a maximum allowable power output value for
the energy storage device, wherein the maximum allowable power
output value decreases as a state of charge of the energy storage
device decreases.
15. The system of claim 14 wherein the processor is further
configured to calculate a second historical power usage value for
the site, the calculating of the second historical power usage
value comprising: for each of a plurality of measurement time
intervals of the instantaneous power usage over an immediately
preceding time window, computing an average amount of power
imported at the site for the measurement time interval; and
selecting the highest average amount across the plurality of
measurement time intervals.
16. The system of claim 15 wherein the processor is further
configured to: if the second historical power usage value is
greater than the target peak value, discharge the energy storage
device in accordance with either: the maximum allowable power
output value; or a power value that is proportional to a maximum
of, for each measurement time interval in the plurality of
measurement time intervals: a product of a time range of the
measurement interval and the difference between an average demand
for the time measurement interval and the target peak value; and if
the second historical power usage value is less than or equal to
the target peak value, discharge or charge the energy storage
device in accordance with the lesser of: the maximum allowable
power output value; and a current power load of the site minus the
target peak value.
17. A non-transitory computer readable storage medium having stored
thereon program code executable by a processor, the program code
comprising: code that causes the processor to monitor instantaneous
power usage at the site, the instantaneous power usage
corresponding to power that is instantaneously imported or exported
at a PCC between the site and a utility-managed energy grid; code
that causes the processor to calculate a first historical power
usage value for the site based on the monitored instantaneous power
usage; code that causes the processor to compare the first
historical power usage value with a target peak value plus a buffer
value; and if the first historical power usage value exceeds the
target peak value plus the buffer value, code that causes the
processor to set the target peak value to the first historical
power usage value.
18. The non-transitory computer readable storage medium of claim 17
wherein the program code further comprises code that causes the
processor to calculate a maximum allowable power output value for
the energy storage device, wherein the maximum allowable power
output value decreases as a state of charge of the energy storage
device decreases.
19. The non-transitory computer readable storage medium of claim 18
wherein the program code further comprises code that causes the
processor to calculate a second historical power usage value for
the site, the calculating of the second historical power usage
value comprising: for each of a plurality of measurement time
intervals of the instantaneous power usage over an immediately
preceding time window, computing an average amount of power
imported at the site for the measurement time interval; and
selecting the highest average amount across the plurality of
measurement time intervals.
20. The non-transitory computer readable storage medium of claim 19
wherein the program code further comprises code that causes the
processor to: if the second historical power usage value is greater
than the target peak value, discharge the energy storage device in
accordance with either: the maximum allowable power output value;
or a power value that is proportional to a maximum of, for each
measurement time interval in the plurality of measurement time
intervals: a product of a time range of the measurement interval
and the difference between an average demand for the time
measurement interval and the target peak value; and if the second
historical power usage value is less than or equal to the target
peak value, discharge or charge the energy storage device in
accordance with the lesser of: the maximum allowable power output
value; and a current power load of the site minus the target peak
value.
Description
BACKGROUND
[0001] Some electric utility customers, most commonly commercial
and industrial customers, are billed two separate charges on their
electricity service bill--a consumption charge and a peak demand
charge. The consumption charge reflects the total amount of energy
that the customer uses over the billing period. For example, if the
customer's site consumes 1000 kilowatt-hours (kWh) at a cost of
$0.07 per kWh, the customer's consumption charge will be
1000.times.$0.07, or $70. In contrast, the peak demand charge
reflects the highest, or peak, amount of power demanded by the
customer within the billing period. For example, if the customer's
site reaches a peak power demand of 20 kilowatts (kW) at a cost of
$8 per kW, the customer's peak demand charge will be 20.times.$8,
or $160. In practice, utility companies usually average power
demand over recurring "demand intervals" (e.g., every 15 minutes),
and then use the highest demand interval average within the billing
period to calculate the peak demand charge.
[0002] The rationale for a dual consumption/demand billing scheme
is that the amount of power required by each customer over time can
vary widely. For instance, some customers may need large bursts of
electricity on an occasional basis, while others need lesser
amounts constantly. As a result, utility companies must operate and
maintain sufficient generation, transmission and distribution
equipment (e.g., transformers, wires, substations, etc.) at all
times in order to meet potential aggregate power demand during high
demand periods, even if such equipment is under-utilized the rest
of the time. These operational/maintenance costs are passed on to
customers proportionally, based on their peak power requirements,
in the form of the peak demand charge.
[0003] For customers that face a high peak demand charge each
billing cycle, it can be economical to install an onsite energy
storage system (e.g., a battery-based system) that performs "peak
shaving." This means that the energy storage system discharges
energy during intervals of high site load, thereby offsetting
energy consumption from the utility grid and reducing, or shaving,
the site's peak power demand. However, existing algorithms for
controlling the flow of energy to/from such systems to achieve peak
shaving (known as "peak shaving algorithms") generally suffer from
a number of drawbacks. For instance, some peak shaving algorithms
are implemented using complex predictive techniques (e.g., machine
learning, neural networks, etc.) that are computationally expensive
and thus are difficult to deploy/execute without expensive
equipment. Other peak shaving algorithms can provide effective
results in certain well-tested scenarios, but are "unstable" and
thus may do a relatively poor job in reducing peak power demand in
other, more general use cases. Accordingly, it would be desirable
to have an improved peak shaving algorithm that addresses the
deficiencies (e.g., high computational cost, poor robustness, etc.)
of prior art solutions.
SUMMARY
[0004] Techniques for controlling an energy storage device to
reduce peak power demand at a site are provided. In one embodiment,
instantaneous power usage at the site can be monitored, where the
instantaneous power usage corresponds to power that is
instantaneously imported or exported at a point of common coupling
(PCC) between the site and a utility-managed energy grid. A
historical power usage value for the site can then be calculated
based on the monitored instantaneous power usage, and the
historical power usage value can be compared with a target peak
value plus a buffer value. If the historical power usage value
exceeds the target peak value plus the buffer value, the target
peak value can be set to the historical power usage value.
[0005] The following detailed description and accompanying drawings
provide a better understanding of the nature and advantages of
particular embodiments.
BRIEF DESCRIPTION OF DRAWINGS
[0006] FIG. 1 is a simplified block diagram of a system environment
according to an embodiment.
[0007] FIG. 2 is a simplified block diagram of a computer system
according to an embodiment.
[0008] FIG. 3 is a graph depicting the algorithmic feature of
"ratcheting" a target peak value according to an embodiment.
[0009] FIG. 4 is a graph depicting the algorithmic feature of
implementing a "dead band" according to an embodiment.
[0010] FIGS. 5A, 5B, and 6 are graphs depicting the algorithmic
feature of "rolling off" the power output from an energy storage
system according to an embodiment.
[0011] FIGS. 7A and 7B are flowcharts depicting a peak shaving
algorithm according to an embodiment.
[0012] FIG. 8 is a pseudo-code implementation of the algorithm of
FIGS. 7A and 7B according to an embodiment.
DETAILED DESCRIPTION
[0013] In the following description, for purposes of explanation,
numerous examples and details are set forth in order to provide an
understanding of various embodiments. It will be evident, however,
to one skilled in the art that certain embodiments can be practiced
without some of these details, or can be practiced with
modifications or equivalents thereof
1. Introduction
[0014] The present disclosure describes a novel algorithm (referred
to herein as the "CV algorithm") for regulating the discharging and
charging of an energy storage system to shave peak power demand at
a site. In various embodiments, the CV algorithm can incorporate
one or more of the following features (which are discussed in
further detail in the sections that follow): [0015] Ratcheting--Set
a target peak value; if site power demand rises above the target
peak value (after offset from the energy storage system), "ratchet
up" the target peak value to meet the new site power demand [0016]
Dead band--Maintain a buffer, or "dead band," above the target peak
value; only raise the target peak value if the site power demand
rises above the target peak value plus the dead band [0017]
Roll-off--As the energy storage system becomes more depleted,
reduce the power output from the energy storage system to extend
its runtime [0018] Dispatch calculation--Discharge increments of
energy from the energy storage system to offset a historical
rolling demand average (based on a utility-defined demand interval)
when the average is above the target peak value
[0019] Taken together, the foregoing features allow for effective
and efficient peak shaving, without incurring the disadvantages or
drawbacks of prior art approaches. For example, by implementing the
ratcheting, dead band, roll-off, and/or dispatch calculation
features, the CV algorithm can determine when (and to what extent)
to discharge the energy storage system in order to reduce the
site's peak power demand for a billing period, without requiring
any predictive foresight of what the site's power load over the
billing period will be. Instead, the CV algorithm can operate
solely based on current and short-term historical information. As a
result, the algorithm is significantly less complex than prior art
predictive algorithms that rely on, e.g., machine learning, neural
networks, etc., and thus can be run on low-cost/commodity
hardware.
[0020] Further, due to its relative simplicity, the CV algorithm
can be easily analyzed for performance/correctness, and can provide
relatively good shaving results under a wide range of conditions.
In certain embodiments, the CV algorithm can be combined with other
algorithms in a layered, or hybrid, approach. For instance, an
alternative peak shaving algorithm (e.g., a neural network-based
algorithm) can be used to make an initial decision regarding
whether, and how much, to discharge the energy storage system. The
CV algorithm can then be applied in view of that initial decision
(as, e.g., a set of constraints) to arrive at a final decision. In
this manner, the CV algorithm can be used as a "backstop" to catch
and correct poor decisions made by more complex, but possibly more
unstable, algorithms.
2. Exemplary System Environment
[0021] FIG. 1 depicts a system environment 100 in which embodiments
of the present invention may be implemented. As shown, system
environment 100 includes a site 102 that comprises infrastructure
(e.g., a meter 104 and a main panel 106) for importing energy from
a utility-managed energy grid 108. The imported energy can be used
to power one or more site loads 110. In one embodiment, meter 104
can be considered a "point of common coupling" (PCC) between energy
grid 108 and site 102. The amount of power imported at the PCC at
any given time is referred to as the site's instantaneous power
demand.
[0022] Site 102 also includes an energy storage system 112
comprising a battery device 114 and a battery inverter/charger 116.
As discussed in the Background section, energy storage system 112
can be leveraged to perform peak shaving--in other words, battery
inverter/charger 116 can discharge (i.e., dispatch) stored energy
from battery device 114 in order to offset site 102's power demand
during high demand periods. Furthermore, battery inverter/charger
116 can use energy that is imported from energy grid 108 to charge
battery device 114 at times when site loads 110 are relatively low.
Through this process, energy storage system 112 can potentially
reduce the peak demand charge billed to the owner of site 102 each
billing cycle.
[0023] To control its operation, energy storage system 112 can be
communicatively coupled with a local controller, such as local
computer 118. Local computer 118 can, among other things, execute
one or more peak shaving algorithms (such as the CV algorithm
described herein) for regulating the discharging/charging behavior
of battery inverter/charger 116. Alternatively or in addition,
energy storage system 112 can be communicatively coupled with an
offsite controller, such as remote computer 120, via a network 122.
In these embodiments, remote computer 120 can take over the duties
of peak shaving calculation. In a particular embodiment, remote
computer 120 can be configured to simultaneously execute peak
shaving algorithms for a fleet of energy storage systems
distributed at multiple sites.
[0024] It should be appreciated that system environment 100 is
illustrative and not intended to limit embodiments of the present
invention. For instance, although energy storage system 112 is
depicted as a battery-based system, other types of energy storage
technologies (e.g., compressed air, flywheels, pumped hydro,
superconducting magnetic energy storage (SMES), etc.) may be used.
Further, the various entities depicted in system environment 100
can have other capabilities or include other
components/subcomponents that are not specifically described. For
example, in certain embodiments, site 102 can include an energy
generation system (e.g., a photovoltaic (PV) system) that is
coupled with energy storage system 112. One of ordinary skill in
the art will recognize many variations, modifications, and
alternatives.
3. Exemplary Computer System
[0025] FIG. 2 depicts a computer system 200 according to an
embodiment. Computer system 200 can be used to implement any of the
computer systems/devices (e.g., local computer 118 or remote
computer 120) described with respect to FIG. 1. As shown in FIG. 2,
computer system 200 can include one or more processors 202 that
communicate with a number of peripheral devices via a bus subsystem
204. These peripheral devices can include a storage subsystem 206
(comprising a memory subsystem 208 and a file storage subsystem
210), user interface input devices 212, user interface output
devices 214, and a network interface subsystem 216.
[0026] Bus subsystem 204 can provide a mechanism for letting the
various components and subsystems of computer system 200
communicate with each other as intended. Although bus subsystem 204
is shown schematically as a single bus, alternative embodiments of
bus subsystem 204 can utilize multiple buses.
[0027] Network interface subsystem 216 can serve as an interface
for communicating data between computer system 200 and other
computer systems or networks (e.g., network 122 of FIG. 1).
Embodiments of network interface subsystem 216 can include wired
interfaces (e.g., Ethernet, CAN, RS232, RS485, etc.) or wireless
interfaces (e.g., ZigBee, Wi-Fi, cellular, etc.).
[0028] User interface input devices 212 can include a keyboard,
pointing devices (e.g., mouse, trackball, touchpad, etc.), a
scanner, a barcode scanner, a touch-screen incorporated into a
display, audio input devices (e.g., voice recognition systems,
microphones, etc.), and other types of input devices. In general,
use of the term "input device" is intended to include all possible
types of devices and mechanisms for inputting information into
computer system 200.
[0029] User interface output devices 214 can include a display
subsystem, a printer, a fax machine, or non-visual displays such as
audio output devices, etc. The display subsystem can be a cathode
ray tube (CRT), a flat-panel device such as a liquid crystal
display (LCD), or a projection device. In general, use of the term
"output device" is intended to include all possible types of
devices and mechanisms for outputting information from computer
system 200.
[0030] Storage subsystem 206 can include a memory subsystem 208 and
a file/disk storage subsystem 210. Subsystems 208 and 210 represent
non-transitory computer-readable storage media that can store
program code and/or data that provide the functionality of
embodiments of the present invention.
[0031] Memory subsystem 208 can include a number of memories
including a main random access memory (RAM) 218 for storage of
instructions and data during program execution and a read-only
memory (ROM) 220 in which fixed instructions are stored. File
storage subsystem 210 can provide persistent (i.e., non-volatile)
storage for program and data files, and can include a magnetic or
solid-state hard disk drive, an optical drive along with associated
removable media (e.g., CD-ROM, DVD, Blu-Ray, etc.), a removable
flash memory-based drive or card, and/or other types of storage
media known in the art.
[0032] It should be appreciated that computer system 200 is
illustrative and not intended to limit embodiments of the present
invention. Many other configurations having more or fewer
components than computer system 200 are possible.
4. Algorithmic Features
[0033] As noted in the Introduction section, in various embodiments
the CV algorithm can incorporate a combination of four
features--ratcheting, dead band, roll-off, and dispatch
calculation--that enable effective and efficient peak shaving.
These features are described in the subsections below.
4.1 Ratcheting
[0034] At a high level, ratcheting involves the following series of
steps: [0035] 1. Initialize a "target peak value" at the start of a
billing period [0036] 2. If site power demand rises above the
target peak value, discharge the energy storage system to offset
the power being imported from the energy grid [0037] 3. If the
discharging of the energy storage system cannot reduce site power
demand down to the target peak value, "ratchet up" the target peak
value to meet the new site power demand [0038] 4. Return to step
(2); loop until the end of the billing period
[0039] To better understand how this process works, consider graph
300 of FIG. 3, which illustrates an exemplary site load curve 302
and target peak curve 304 mapped against power (on the Y axis) and
time (on the X axis). As shown, target peak curve 304 is
initialized to a value of 50 kW at time A. Further, site load curve
302 (which is equivalent to site power demand when the energy
storage system is not being discharged) starts at 40 kW at time A
and gradually increases. The site load curve may represent an
instantaneous power demand or the average power demand over a
period of time (for example, 15 minutes).
[0040] At time B, site load curve 302 crosses the current target
peak value of 50 kW. As a result, the energy storage system begins
discharging energy to offset the power imported from the energy
grid per step (2) above. This offset is depicted via arrows which
show the site's power demand (which was previously equivalent to
the site power load) being reduced down to the current target peak
value (50 kW).
[0041] After time B, site load curve 302 continues to rise. At time
C, site load curve 302 has risen to a point where the energy
storage system does not have enough power to reduce site power
demand down to the current target peak value of 50 kW. As a result,
target peak curve 304 is ratcheted up to the new site power demand
of (site load minus energy storage system offset) per step (3)
above.
[0042] This ratcheting continues until target peak curve reaches 70
kW at time D, at which point site load curve 302 begins to fall
(thereby allowing the energy storage system to fully offset site
power demand down to the new target peak value of 70 kW). Target
peak curve 304 remains at 70 kW for the remainder of this example
(and will continue to stay at this level unless site load curve 302
again rises to a point where the energy storage system cannot fully
offset site power demand down to 70 kW).
[0043] With the ratcheting shown in FIG. 3, the energy storage
system can avoid discharging its energy source unless such
discharging will reduce the highest power demand encountered so far
during the billing period (as embodied by the target peak value).
Accordingly, this increases the likelihood that the energy storage
system will be able to shave the peak power demand for the billing
period, without requiring prior knowledge of how site load curve
302 will behave. In one set of embodiments, the target peak
value/curve can be compared against the site's instantaneous power
demand (i.e., instantaneous power draw at the PCC) in order to
determine when to discharge the energy storage system and when to
ratchet the target peak value. In other embodiments, the target
peak value/curve can be compared against a historical power
usage/demand value (e.g., a demand average over a historical
window). This alternative is discussed in further detail below.
4.2 Dead Band
[0044] The dead band feature is closely related to ratcheting and
pertains to the manner in which the target peak value is ratcheted
up to a higher value. When this feature is implemented, the target
peak value is not raised immediately upon determining that the
energy storage system cannot fully offset site power demand down to
the current target peak (per step (3) above). Instead, a relatively
small dead band, or buffer, is maintained above the target peak
value, and the target peak value is only ratcheted when the site
power demand exceeds the current target peak value plus the dead
band.
[0045] This process is illustrated in graph 400 of FIG. 4, which
depicts the same site load curve 302 of FIG. 3, as well as a dead
band 402 above a new target peak curve 404. In the example of FIG.
4, target peak curve 404 remains flat at time C (i.e., the time at
which target peak curve 304 is ratcheted in FIG. 3) because while
the site power demand (406) has begun to rise, it has not yet
exceeded the current target peak value plus dead band 402. At time
E, site power demand 404 finally rises above the target peak value
plus dead band 402, thereby causing target peak curve 404 to be
ratcheted upward to match the new site power demand.
[0046] One reason why the dead band feature is a useful improvement
over standard ratcheting is that there is typically some lag time
between an actual increase in site power demand and the ability of
the energy storage system to begin discharging. This may cause the
standard ratcheting process to increase the target peak value even
though the energy storage system is, in fact, capable of offsetting
power demand down to the current target peak value (but simply has
not responded quickly enough). If this scenario occurs multiple
times over the course of a billing period, the target peak may be
incorrectly raised above the actual peak demand for the period.
Implementing a dead band as described above can avoid this since
the energy storage system is given some time to "catch up" to the
current site power demand before the target peak value is
ratcheted.
[0047] In one embodiment, the size of the dead band can be
determined empirically before the CV algorithm is executed at a
particular site. For instance, a historical load curve for the site
can be determined, and a number of simulations of the CV algorithm
can be executed with respect to the historical load curve, with
different dead band sizes. The dead band size that produces the
best shaving results can then be selected for use during actual
runs.
[0048] In alternative embodiments, the dead band may be dynamically
determined/adjusted in real-time while the CV algorithm is executed
on "live" demand/load data at the site.
4.3 Roll Off
[0049] Roll-off refers to the notion of adjusting the amount of
power that is allowed to be discharged from the energy storage
system based on its state of charge (SoC). Generally speaking, if
the SoC is higher, a greater amount of power will be allowed to be
discharged, and if the SoC is lower, a lesser amount of power will
be allowed to be discharged. This effectively extends the runtime
of the energy storage system and increases the probability that the
energy storage system will have energy available to shave the
site's peak power demand for a billing period when that peak demand
actually occurs.
[0050] To illustrate this, FIG. 5A is a graph 500 that depicts an
exemplary site load curve 502 and site demand curve 504 when the
roll-off feature is not implemented. As shown, site load curve 502
and site demand curve 504 start out at the same level at time A. At
time B, site load curve 502 begins to rise and the energy storage
system is discharged at full power, thereby causing site demand
curve 504 to rise at a slower pace since the site demand is now
being offset by the energy storage system. Unfortunately, at time
C, the energy storage system runs out of energy. As a result, site
demand curve 504 jumps upward to match site load curve 502, and the
energy storage system is ultimately unable to shave the peak demand
for the billing period (which occurs at time D).
[0051] In contrast, FIG. 5B is a graph 520 that depicts the same
site load curve 502 and a new site demand curve 522 when the
roll-off feature is implemented. Like FIG. 5A, at time B of FIG.
5B, the energy storage system is discharged at full power in order
to offset site load curve 502. However, rather than continuing to
discharge at full power, the energy storage system gradually
reduces, or rolls off, its output power over time (shown by the
gradually reducing offset). This enables the energy storage system
to extend its runtime long enough to offset the peak load/demand
that occurs at time D.
[0052] The particular manner in which the energy storage system
rolls off its allowable output power can be defined as a function
of the system's SoC as depicted in graph 600 of FIG. 6. In the
example of FIG. 6, roll-off function 602 stays at full output power
(i.e., the maximum rated power for the energy storage system) from
full charge (i.e., 100% SoC) to a predefined SoC threshold 604. The
allowable output power then falls off at an exponential rate until
the energy storage system is fully depleted (i.e., reaches 0% SoC).
With this type of roll-off function, SoC threshold 604 can be
considered to control the "aggressiveness" of the algorithm. For
instance, a lower SoC threshold will cause the energy storage
system to more aggressively shave site power demand at the start,
but will cause the system to deplete faster (and thus runs the risk
of running out of energy if the site power demand continues to
rise). On the other hand, a higher SoC threshold will not shave as
much demand at the start, but will cause the energy storage system
to last for a longer time during an extended continuously rising
"hill" of demand (and thus may more effectively reduce the peak
demand in such a scenario).
[0053] In other embodiments, alternative roll-functions that are
different from roll-off function 602 of FIG. 6 may be used. For
example, one such alternative function may decline in a linear
fashion after the SoC threshold is reached. Another alternative
function may not have any SoC threshold at all, or may have
multiple SoC thresholds that define boundaries between multiple
sub-functions within the roll-off function. One of ordinary skill
in the art will recognize many variations, modifications, and
alternatives.
4.4 Dispatch Calculation
[0054] The dispatch calculation feature builds upon aspects of the
previous three concepts and refines the manner in which the CV
algorithm determines when to discharge the energy storage system in
order to maximally reduce the peak demand charge. Recall that
utilities typically do not calculate the peak demand charge based
on the highest instantaneous power demand within a billing period;
instead, they determine average power demand over recurring demand
intervals (e.g., every 15 minutes), and then use the highest demand
interval average for the billing period as the site's peak
demand.
[0055] This means that instantaneous power demand will not always
be the best criterion for determining when to discharge the energy
storage system. For instance, consider a 15 minute demand interval
where instantaneous power demand rises high (e.g., above the target
peak value) for the first 5 minutes, and then falls below the
target peak value. In this case, if the energy storage system were
controlled solely based on instantaneous power demand, the system
would stop discharging after 5 minutes. However, it may be the case
that the average power demand for the current demand interval is
higher than any previous 15 minute average (due to, e.g., the sharp
rise within the first 5 minutes). Accordingly, it may be beneficial
to continue discharging the system further into the demand interval
(even if instantaneous power demand has fallen below the target
peak value) in order to bring the overall demand interval average
down.
[0056] To account for the foregoing scenario (and other similar
scenarios), the dispatch calculation feature constantly attempts to
lower a power demand value that is based on a historical, rolling
window, where the window is based on the utility-defined demand
interval.
[0057] This increases the likelihood that the highest demand
interval average for the billing period (rather than simply the
highest instantaneous power demand) will be shaved.
[0058] In a particular embodiment, when this feature is enabled,
the CV algorithm can determine an average power demand value for
each of a number of measurement intervals, where each measurement
interval corresponds to a window from the current time to a past
time within the immediately preceding demand interval. By way of
example, if the demand interval is 15 minutes, the CV algorithm may
calculate an average power demand value for a first interval from
the current time back 1 minute, a second interval from the current
time back 2 minutes, a third interval from the current time back 3
minutes, and so on up to a fifteenth interval from the current time
back 15 minutes, resulting in 15 separate averages. The CV
algorithm can then check whether any of these averages is greater
than the current target peak value; if so, the energy storage
system can be discharged. In one embodiment, the energy storage
system can be discharged at its maximum allowable power until the
next algorithmic time step (per the roll-off function described in
section 4.3 above). In another embodiment, the energy storage
system can be discharged at a power that is proportional to the
maximum value over all intervals of the product of a measurement
interval time, and the difference between the interval average
demand and the target peak . This can be represented as:
P=k.times.max((IntervalDemand(i)-TargetPeak).times.TimeOfAveragingInterv-
al(i)
In the equation above, k is a constant, IntervalDemand(i) is the
average demand computed for the i-th measurement interval,
TargetPeak is the target peak value, and TimeOfAveragingInterval(i)
is the averaging time used for interval i.
[0059] In other embodiments, the energy storage system can be
discharged to the extent needed to bring the maximum demand
interval average down to the current target peak value.
[0060] If none of the calculated averages are greater than the
target peak value, the energy storage system can either be charged
or discharged based on the site's instantaneous power demand.
5. The CV Algorithm
[0061] With the foregoing features in mind, FIGS. 7A and 7B depict
a flowchart 700 of the CV algorithm according to an embodiment. In
the following description, flowchart 700 is described as being
executed by local computer 118 for the purpose of controlling
energy storage system 112 of FIG. 1. However, flowchart 700 may
also be executed by remote computer 120 or any other computing
device that is in communication with system 112.
[0062] Starting with FIG. 7A, at block 701, local computer 118 can
enter a loop for each billing period. At block 702, local computer
118 can initialize a target peak value and a dead band (i.e.,
buffer) value for the current billing period. As noted previously,
these values can be determined empirically or via a dynamic
mechanism.
[0063] Upon initializing the target peak and dead band values,
local computer 118 can enter a loop for each algorithmic time step
within the billing period (block 704). Since the CV algorithm is an
iterative algorithm, these time steps reflect the intervals at
which local computer 118 executes the iterations of the algorithm.
Each time step can be set to, e.g., 10 seconds, 30 seconds, 1
minute, 5 minutes, or any other user-defined value.
[0064] Within the loop of block 704, local computer 118 can first
determine the current instantaneous power usage, or demand, at the
site's PCC and calculate a historical average of the instantaneous
power usage over a preceding utility-defined demand interval
(blocks 706 and 708). For example, if the demand interval is 15
minutes, local computer 118 can calculate a historical average of
instantaneous power usage over the past 15 minutes. Local computer
118 can then compare the historical average calculated at block 708
with the current target peak value plus dead band value (block
710). If the target peak value plus dead band value is exceeded,
local computer 118 can ratchet up the target peak value to match
the historical average (block 712).
[0065] Further, at block 714, local computer 118 can calculate a
maximum amount of power that can be discharged from energy storage
system 112 (i.e., "max allowable power") based on system 112's
current SoC. In one embodiment, this can comprise invoking a
roll-off function as described in section 4.3 above.
[0066] Turning now to FIG. 7B, at block 716, local computer 118 can
calculate historical averages of instantaneous power usage/demand
over N measurement intervals in the preceding demand interval. For
instance, per the example noted in section 4.4, if the demand
interval is 15 minutes and each algorithmic time step is 1 minute,
local computer 118 can calculate 15 different averages, where each
average reflects average power usage from the current time back N
minutes (N=1 to 15).
[0067] Once these historical averages are computed, local computer
118 can determine whether the largest, or max, historical average
exceeds the target peak value. If so, local computer 118 can
instruct energy storage system 112 to discharge energy at the max
allowable power determined at block 714 (block 720) and the current
loop iteration can end (block 728).
[0068] If not, local computer 118 can move on to comparing the
current site load versus the target peak value (block 722). If the
current site load exceeds the target peak value, local computer 118
can instruct energy storage system 112 to discharge energy to the
extent required to bring the site load down below the target peak
value (block 724); otherwise, local computer 118 can instruct
energy storage system 112 to charge itself (block 726). The current
time step loop iteration can subsequently end (block 728) and the
loop of block 704 can repeat until the end of the billing period is
reached (block 730). The entire process can then be iterated for
subsequent billing periods.
[0069] To further illustrate the operation of the CV algorithm,
FIG. 8 depicts an exemplary pseudo-code implementation 800 of the
algorithm according to an embodiment. The variables used in this
implementation are defined as follows: [0070] P.sub.target--Target
peak value [0071] P.sub.deadband--Dead band value [0072]
P.sub.ppc--Rolling demand interval average of instantaneous power
usage/demand at the PPC [0073] P.sub.shavemax--Maximum allowable
power of the energy storage system given its current SoC [0074]
P.sub.shavemaxrated--Maximum rated power of the energy storage
system [0075] E.sub.storage--Total amount of stored energy in the
energy storage system [0076] T.sub.runtime--Target runtime of the
energy storage system [0077]
Max.sub.demandintervalP.sub.ppc--Maximum demand average over N
measurement intervals within the immediately preceding demand
interval; in one embodiment, this variable can be calculated as
follows:
[0077] max ( i = now n E i .DELTA. T now - n ) ##EQU00001## where E
n = P n .times. .DELTA. t n ##EQU00001.2## [0078] In the equations
above, P.sub.n corresponds to the instantaneous power usage/demand
at measurement interval n and .DELTA.t.sub.n corresponds to the
length of measurement interval n. Thus, E.sub.n is the total amount
of energy consumed during interval n and
Max.sub.demandintervalP.sub.ppc is the max power usage across all
values of n (i.e., 1 to N). [0079] P.sub.shave--Instantaneous power
output of the energy storage system; positive values represent
power output, and negative values represent charging (if the system
is capable of charging) [0080] P.sub.load--Power of the site load,
not including offset from the energy storage system
[0081] At line 802 of listing 800, a FOR loop is initiated for each
time step of the current billing period. At line 804, the target
peak value is calculated as the maximum of (current target peak
value plus dead band value) and the maximum rolling demand interval
average of power usage/demand. Thus, the target peak value is
ratcheted up if the maximum rolling demand interval average exceeds
the current target peak plus dead band.
[0082] At line 806, the maximum allowable power from the energy
storage system is calculated as the minimum of the maximum rated
power of the system and the current storage level divided by target
runtime. This reduces the allowable power in a linear fashion as
the SoC of the energy storage system decreases.
[0083] Finally, at lines 808-814, the maximum demand average over N
measurement intervals in the preceding demand interval is
calculated and compared to the target peak value. If the maximum
average exceeds the target peak value, the energy storage system is
discharged at its maximum allowable power. Otherwise, the power
output from the energy storage system is calculated as the minimum
of the maximum allowable power and the difference between the
current site load and the target peak value. Note that, if the site
load is greater than the target peak value, the power output
calculated at line 814 will be a positive value, which means the
energy storage system will be discharged. On the other hand, if the
site load is less than the target peak value, the power output may
be a negative value, which means the energy storage system may be
charged. In this case, the amount of charge will correspond to the
difference between P.sub.load and P.sub.target, which ensures that
the charging will not cause site power demand to exceed the target
peak value.
[0084] The above description illustrates various embodiments of the
present invention, along with examples of how aspects of the
embodiments may be implemented. These examples and embodiments
should not be deemed to be the only embodiments, and are presented
to illustrate the flexibility and advantages of the present
invention as defined by the following claims. For example, although
certain embodiments have been described with respect to particular
flowcharts and steps, it should be apparent to those skilled in the
art that the scope of the present invention is not strictly limited
to the described flowcharts and steps. Steps described as
sequential may be executed in parallel, order of steps may be
varied, and steps may be modified, combined, added, or omitted. As
another example, although certain embodiments have been described
using a particular combination of hardware and software, it should
be recognized that other combinations of hardware and software are
possible, and that specific operations described as being
implemented in software can also be implemented in hardware and
vice versa.
[0085] The specification and drawings are, accordingly, to be
regarded in an illustrative rather than restrictive sense. Other
arrangements, embodiments, implementations and equivalents will be
evident to those skilled in the art and may be employed without
departing from the spirit and scope of the invention as set forth
in the following claims.
* * * * *